Are 1nm chips the holy grail of computing, or is it possible to make chips with fractional electrons without floating point errors?
You clearly don't understand much about tech.The answer is that CPUs already aren't quantum coherent so it seems we can in fact go smaller.
>>107990864Technology peaked at 7nm. Its refinement (Called 6nm sometimes) is all you need.
>>107990864It's possible and it will make many things obsolete. If you can split an electron into two, that would mean the same number of them would be able to represent more data, so things like filesizes, become meaningless. Although there isn't any hard limit to how small you can make them, there are practical reasons to consider. If they're too small, they're more likely to escape into the air so data loss becomes a huge problem.
Quantum computing is not really a meme
>>107990864There are more ways to store or compute data than just flash cells and transistors, you limited homosapien.
>>107990962Nobody cares about outdated technology. Electrons by their very nature, travel the shortest available path, which is why high-speed read/write is even possible.
>>107990864they lie, we are still using 20nm process, our transistors aren't much smaller than they were, they just found tricks to pack more and then make some retarded surface / transistor to come up with some retarded transistor size.
>>107990984>Electrons, travel the shortest available pathlol, if only that was actualy true
>is it possible to make chips with fractional electronsThey still haven't figured out how to convert electrons from wave to particle and vice-versa. Until they can do that, we're stuck with particle-based chips where 1 electron = 1 bit.
>>107991005Go read a book about Om's law.
>>107990864Usecase for smaller electrons?
>>107991028>abstractions are what's realy happeningyou got a midwit understanding of electrons.
>>107991134>>107991028add to that, at the scale of processors, those abstractions are no longer good enough, especialy with bleeding edge stuff.
>>107991134The only way an abstraction can work is when there's a physical process underneath. Your data does not sit in a vaccum. They are actual electrons.Your inability to understand basic concepts, is quite telling.
>>107991152you are missing the point entirely, my point is that "electron follow the shortest path" is an useful abstraction for simple electronics, but it isn't true whatsoever, especialy at smaller scales, and even when doing normal electronics, because it isn't true there are exceptions to take into account.it's an useful abstraction, but it isn't generaly true.same thing with newton's gravitation laws, they are still useful and used today, they are an useful abstraction, but they are fundamentaly wrong and will break in other cases, that's why we use GR when it's necessary, GR didn't mean that we suddenly stopped using newton's equation, because they are simpler and still more than good enough for many uses, it's not because an abstraction is wrong that it cannot be useful, but you have to understand the limitation of said abstraction and be ready to ditch it when it is no longer useful for your case.anyway, if you knew the smallest amount of QM you'd know, no, electrons do not follow the shortest path, they *may* if you make an average path, but even then that's not a guarantee.
>>107991152>>107991217also, in many cases treating them as particles with a fixed point is useful and good enough, but again, that's a false abstraction, it is useful but you shouldn't confuse the abstraction for the real thing, which is a lot more complex and for many case, you have to deal with the complexity because the abstraction is no longer good enough.
>>107991152>>107991217>>107991226CPU engineering being one of those cases where the "particle" abstraction is hardly good enough anymore, ie tuneling.
>>10799086414nm was unironically peak node.It was easier to cool, yet still incredibly power efficient at lower loads.
Because the location of an electron is actually a distribution of probability, quantum tunneling means there's a theoretical minimum feature size (10 angstroms if I recall of the top of my head) after which you literally cannot get smaller without having to treat your gates like qubits. Plus as you get smaller there's electromigration, where atoms get moved around by electrons which fucks up your metal layers.We still have a way to go before we get cucked by quantum physics though (backside power delivery, thermal vias, etc.)
>>107991253The probability distribution applies only to wave-like electrons. For things like memory and SSDs where you need bit-perfect data, they convert electrons to particles.
>>107991280what the fuck are you talking about nigga the electrons don't care what you name the IP block
>>107991287If they used wave-like electrons for data, it would get corrupted, you fucking retard. When you copy a file, do you want the actual file or just an approximation of it?Particle-like electroncs can be used for bit-perfect mapping of 1 electron = 1 bit of information.
>>107991299
absolutely shit thread >>107990864>1nm chips why not 0.5nm chips? there is no holy grail >fractional electrons these dont exist. if you split an electron (im not even sure anyone has done this) its no longer an electron. its also impossible to guess how this half electron will behave >>107990914>the same number of them would be able to represent more datahow? what data is being represented by a single electron? there are no electron states to read, only if its there or if it isnt by your logic one electron can represent infinite data because you have spaces where there arent electrons that all have a zero written in them >things like filesizes [will] become meaninglessimpossible, its just impossible to have infinite filesize >>107990984they dont travel the shortest path, they travel the easiest path>>107991086ask a particle physicist, but the usecase is probably just research >>107991217>newton's gravitation lawsif by mentioning this you mean that we dont have a complete model of the universe youre delusional. a complete model isnt needed, and newtons laws havent been disproven. they are still used and they arent fundamentally wrong. general relativity measures something different this is like saying that a ruler is fundementally wrong and breaks when you try use it to measure time and thats why a stopwatch is better. finding an all encompassing model of the universe means finding a tool that does the same job as a ruler and a stopwatch>>107991226meaningless, reword this using simpler language >>107991231>he thinks cpu transistors are small enough that tunneling is a concern >>107991253>actually a distribution of probabilityit isnt in reality, but because the tools to exactly pinpoint an electron in real time dont exist people just estimate where it probably isthe rest of your post has nothing to do with quantum physics (i bet you have no idea what quantum even means) >>107991280holy kek
>>107991607>why not 0.5nm chips?Because at 0.5nm, you would have to split an electron into two and that can potentially corrupt data.
>>107990864smaller transistors that are more densely packed on a die ultimately result in a decrease in performance and efficiency for the end user. We've been seeing hardware perform worse and worse the last several years despite "advancements" being made. It's not just Moore's law slowing down, but reversing. I call it Jeet's law.>Every two years, the amount of Indians producing poor computer code doubles.>>107991242trvke
>>107990864nm doesn't mean anything, it's a marketing term
1nm doesn't matter when hardware is bloated due to X86 and software is also bloated due to jeetmade spaghetti code, frameworks and obfuscated high level languages
